In this essay, I define art based on literary works, critical theory, and modern philosophy. I then argue against the validity of AI generated art, using formal creative venues as a frame of reference.
As an artist, I have experienced a spectrum of emotions following the advent of generative artificial intelligence (referred to simply as AI for the rest of this paper), from anger and defiance to grief and acceptance. This paper explores a definition of creation first from a literary perspective, then expands to encompass different art forms, and finally discusses generative artificial intelligence’s role in the creative process. AI can be a powerful tool for soliciting instant and unlimited feedback, but leaning on it to create dilutes artistry and contributes to a disconnected society. Thus, outlets such as art competitions, award shows, and literary journals should not allow AI-generated content in any portion of a creative work.
To create a unified definition of authorship, we can consider previous works of critical theory along with anecdotal evidence. Dan Sinykin sums up the author as existing in three modalities: as a vessel through which external inspiration manifests, as language speaking for itself, and as “social acts”. In all three definitions, the author acts as a conduit, channeling the collective through the individual. This definition can extend beyond just the author to encompass different forms of creative expression, too.
However, this definition does not identify the purpose behind creation. Multiple renowned authors have touched on the root of artistry in their works. In the first part of her memoir, Childhood, Tove Ditlevsen writes, “Even though no one else cares for my poems, I have to write them because it dulls the sorrow and longing in my heart” (92). Reflecting on his short stories in The Things They Carried, war veteran Tim O’Brien writes, “What stories can do, I guess, is make things present. I can look at things I never looked at. I can attach faces to grief and love and pity and God. I can be brave. I can make myself feel again” (204). And in Letters to a Young Poet, Rainer Maria Rilke advises the young poet Franz Kappus that “A work of art is good if it has arisen out of necessity” (5). From these accounts, art manifests as a necessary act of expressing oneself and reconciling with life’s hardships. Through writing, an author can channel their thoughts, feelings, obsessions, and questions into a concrete medium that communicates meaning beyond the words themselves. This act allows the collective — inspiration, language, experiences, materiality, culture, and so on — to merge and shine through the individual’s voice. For an artist, creation is inevitable. It is a necessity.
Though the explanations of art as channeling the collective through the individual and as a necessity encompass the human aspect of creation, they remain unquantifiable. To ground the definitions, we can consider Ted Chiang’s claim in his article "Why A.I. Isn’t Going to Make Art" that “art is something that results from making a lot of choices.” These choices form an artist’s style and voice. Every word, every brush stroke, every chord — each of these are meticulously constructed with intent. By this definition, art takes shape as any form of expression a creator has put work into. Thus, we arrive at our final definition of art: the necessary act of making deliberate choices to channel the collective through the individual.
On the surface, the idea of artistry seems to have been shaken up by generative AI’s public release. A plethora of AI-generated assets have since surfaced in all areas, from social media posts and blog articles to graphic design and marketing copy. However, in reality, AI is incapable of true creation. In Chiang’s article, he claims that generative AI tools either “take an average of choices that other writers have made,” which amounts “to the least interesting choices possible,” or mimic the style of an existing author. No matter the method, AI plagiarizes by non-consensually using words and ideas from its dubiously acquired training data as its own. Though it does not necessarily intend to plagiarize, it does so simply by design. This concept extends to visual and auditory sources as well. Additionally, the LLMs output will far exceed the user’s input, and the user is spared from having to make energy-consuming decisions. Thus, “the person entering the prompt can’t claim credit for” the AI’s output, and neither can the AI, despite its illusion of originality. Furthermore, Chiang argues that AI cannot use language, as language “requires an intention to communicate” — it “lies on top of these other experiences of subjective feeling and of wanting to communicate that feeling.” The desire to communicate a “subjective feeling” applies to our definition of art as a necessity. As such, AI fundamentally is unable to express itself, whether through language or other media, but people still tend to project feelings onto it due to its common anthropoglossic design with a chat interface (Bergstrom and West). Then, AI, at best, serves as an aggregation tool and, at worst, unjustly plagiarizes from real creators.
With AI as an aggregation tool, there must be a point at which a work is no longer created but instead generated. A work deserves as much attention as the amount of effort that is put into it (Chiang). In other words, a completely AI-generated artwork deserves no attention since it is made on the foundation of stolen, unattributed, and pre-existing choices. Additionally, aggregated data is depersonalized, and this concept applies, too, to AI, a choice-averaging machine. Using any form of generative AI reduces the number of choices an artist has made in their work, leading to communication with muffled intention. The outcome is filtered through training data elements far removed from the creator’s personal experiences and motivations. Thus, by incorporating AI-generated output, artists decrease their connection with their audience. In doing so, they lose parts of their voice and authenticity.
The notion that generative AI can contribute directly to creative works stems from the current landscape of capitalism. The society we currently live in emphasizes efficiency and values the product over the process. French philosopher Baudrillard conceptualizes this type of culture as a consumer society of manufactured desires, where everything is “asepticized into cold, clinical communication” (Reid). Generative AI and LLMs take this “clinical communication” to the extreme by producing endless, meaningless outputs, devoid of intentionality. Creating a work with commercially available generative AI is a means to an end. It generates a product without a personal process, with its output far exceeding the input. It expands upon the skeleton of an idea, auto-completing based on stolen works rather than genuine artistic process. The spark of creation is not nurtured with care but instead thrown into an expedited blender. Generative AI’s existence is a form of digital colonization, where the end product is built on unjustly stolen work, and the colonizers are technologists in power who believe art is a product and not a necessary process.
Allowing commercially available generative AI in formal creative venues sets a precedent and encourages the public to view art as content and not the encapsulation of a human experience. Yet to ban it completely would be a disservice to those who express their creativity by harnessing machines and algorithms. Since we understand art as “the necessary act of making deliberate choices to channel the collective through the individual,” I believe generative AI use should not be used to generate any part of the final work. It may be used to solicit constructive feedback on the work, as long as the process does not undermine the choices made by the artist. Any AI that the creator programs or trains themselves should also be allowed, as the human input exceeds the machine output. Ultimately, more human than machine choices should be made for a piece to truly constitute art. Over-reliance on AI generations dilutes artist voices and distances them from the audience, leading to a disconnected society where human experiences are halfheartedly communicated through a veil of aggregated output.
Works Cited
- Bergstrom, Carl, and Jevin West. “Lesson 1: Autocomplete in Overdrive.” Modern-Day Oracles or Bullshit Machines?, https://thebullshitmachines.com/lesson-1-autocomplete-in-overdrive/. Accessed 12 Mar. 2025.
- ---. “Lesson 3: Turing Tests and Bullshit Benchmarks.” Modern-Day Oracles or Bullshit Machines?, https://thebullshitmachines.com/lesson-3-turing-tests-and-bullshit-benchmarks/. Accessed 11 Mar. 2025.
- Chang, Kent K., et al. Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4. arXiv:2305.00118, arXiv, 20 Oct. 2023. arXiv.org, https://doi.org/10.48550/arXiv.2305.00118.
- Chiang, Ted. “Why A.I. Isn’t Going to Make Art.” The New Yorker, 31 Aug. 2024. www.newyorker.com, https://www.newyorker.com/culture/the-weekend-essay/why-ai-isnt-going-to-make-art.
- Ditlevsen, Tove. “Childhood.” The Copenhagen Trilogy, translated by Tiina Nunnally, Picador, 2021, pp. 1–99.
- Kundera, Milan. The Unbearable Lightness of Being. Translated by Michael Henry Heim, Harper Perennial Modern Classics, 2009.
- Reid, Kelly. “Jean Baudrillard’s Contributions to Semiotic and Structural Studies.” International Journal of Baudrillard Studies, https://baudrillardstudies.ubishops.ca/jean-baudrillards-contributions-to-semiotic-and-structural-studies/. Accessed 7 Mar. 2025.
- Rilke, Rainer Maria. Letters to a Young Poet. Translated by Charlie Louth, Penguin Classics, 2016. Sinykin, Dan. “Conglomerate Authorship.” Big Fiction: How Conglomeration Changed the Publishing Industry and American Literature, Columbia University Press, 2023, pp. 8–12.
- Photo by Ahmad Odeh on Unsplash